Banca de QUALIFICAÇÃO: WENDELL MEDEIROS DE AZEVEDO

Uma banca de QUALIFICAÇÃO de MESTRADO foi cadastrada pelo programa.
STUDENT : WENDELL MEDEIROS DE AZEVEDO
DATE: 26/01/2026
TIME: 10:00
LOCAL: Remoto
TITLE:

Generative Artificial Intelligence in the Public Procurement Planning Phase: a Practical Application in the Federal Court of Rio Grande do Norte

KEY WORDS:

Retrieval-Augmented Generation; Law 14,133/2021; Judiciary; Document Automation; High-Fidelity Prototyping.

PAGES: 158
BIG AREA: Ciências Exatas e da Terra
AREA: Ciência da Computação
SUBÁREA: Metodologia e Técnicas da Computação
SPECIALTY: Sistemas de Informação
SUMMARY:

Law No. 14,133/2021 introduced significantly more rigorous documentary requirements in the planning phase of public procurement, imposing cognitive overload on technical teams and heterogeneity in the quality of documents produced by various administrative units. At the Federal Court of Rio Grande do Norte, analysis of 340 bidding waiver processes conducted between 2017 and 2024 revealed a 57.53% increase in median planning time after the legislative transition, rising from 14.60 to 23 days. This research aims to design the Public Procurement Administrative Documents Elaboration System (SEDAC), an intelligent system based on Retrieval-Augmented Generation technique to simplify and standardize the preparation of documents required in the planning phase. The methodology adopted integrates Design Thinking and Lean Startup across six phases: institutional diagnosis, technicalscientific review, requirements gathering, conceptual modeling, prototyping and validation, and market analysis. The development resulted in a conceptual architecture grounded in a Naive RAG model with curated knowledge bases, comprising legislation, case law, document templates, and normative guidelines, as well as a high-fidelity mockup demonstrating the conceptual feasibility of the solution. The financial viability analysis presented favorable indicators, with a Net Present Value of R$ 687,797.96, Internal Rate of Return of 47.11%, and a 2-year payback period. The research contributes to the literature by demonstrating the practical applicability of RAG architecture in regulated domains of Brazilian public administration, evidencing that the normative complexity of the planning phase can be addressed through technological innovation that preserves final human validation and legal compliance.

COMMITTEE MEMBERS:
Presidente - 2459504 - AQUILES MEDEIROS FILGUEIRA BURLAMAQUI
Interno - 1555898 - DIEGO RODRIGO CABRAL SILVA
Interno - 1943220 - ORIVALDO VIEIRA DE SANTANA JUNIOR
Externo ao Programa - 3410576 - FELIPE MACEDO ZUMBA - UFRN